As a function of a known sound reproducing apparatus mounted on a vehicle, there is known a noise responsive controller for correcting a gain of each frequency so as to listen to music even in a noise environment by using a masking model. In addition, an example of the prior art related to the present application includes Patent Document 1.
However, in a sound reproducing apparatus having the noise responsive controller for correcting a music signal so as not to be masked with a noise with an increase in a noise, when the music signal is corrected on the basis of the known masking model, a correction amount is so considerably increased that a listener can sense what the music signal is being corrected. For this reason, the known correction method results in disturbing a natural music play.
As correction methods of correcting the music signal, there have been contrived three types of correction methods: the correction method of changing a volume itself, the correction method of using a correction filter of each bandwidth in the state where the frequency bandwidth is divided in detail, and the correction method of using an adaptive filter or the like. Even when the correction methods are used, all of the correction methods may disturb a natural music listening at the time of inputting music as an input signal. It is considered that the reason is why as it is clear from a cocktail party effect, humans have an acoustic sense ability to discern a different sound source at the time of listening to some extent. For example, when the adaptive filter with a complicated filter shape is added to the music signal, humans can sense unnaturalness.
Further, in order to use a masking amount as the correction amount, as a specific calculation equation of the masking amount, for example, an experimental equation based on an acoustic psychology is used. However, the experimental equation based on the acoustic psychology is an experimental equation obtained by not a masking model of a music signal from a sound reproducing apparatus, but a masking model of a noise source such as a pure sound, a band noise, or a white noise. For this reason, when the experimental equation based on the acoustic psychology is applied to real music and a noise source, a correction amount becomes particularly large in a low pass in which a large power is concentrated. As a result, a natural music listening is notably disturbed.    [Patent Document 1] Japanese Patent Unexamined Publication No. 7-307632